Add Iterator interface fix #6

[Moelf]

The code is a bit messy but the idea is there. I will clean up a bit before merging just want to know if people have thoughts on the implementation. The speed seems okay, allocation probably can be reduced, but a revamp on the interped_data() would benefit all use case.

It's thread-safe to iterator through multiple branches by simply zipping them together:

julia> rf = ROOTFile("./tree_with_large_array.root")
ROOTFile("./tree_with_large_array.root") with 1 entry and 17 streamers.

julia> it1 = BranchItr(rf, rf["t1/int32_array"])
itBranchItr{Int32, UnROOT.Nojagg}
Branch: int32_array
Total entry: 100000

julia> it2 = BranchItr(rf, rf["t1/float_array"])
BranchItr{Float32, UnROOT.Nojagg}
Branch: float_array
Total entry: 100000

julia> for (a,b) in zip(it1,it2)
           a, b
       end

[codecov]

Codecov Report

Merging #28 (271002f) into master (a975f22) will decrease coverage by 0.07%.
The diff coverage is 79.06%.

❗ Current head 271002f differs from pull request most recent head bd40294. Consider uploading reports for the commit bd40294 to get more accurate results

@@            Coverage Diff             @@
##           master      #28      +/-   ##
==========================================
- Coverage   81.58%   81.50%   -0.08%     
==========================================
  Files           8        9       +1     
  Lines         885      930      +45     
==========================================
+ Hits          722      758      +36     
- Misses        163      172       +9     

Impacted Files	Coverage Δ
src/UnROOT.jl	100.00% <ø> (ø)
src/arrayapi.jl	74.46% <74.46%> (ø)
src/bootstrap.jl	79.56% <80.00%> (+<0.01%)	⬆️
src/streamers.jl	88.88% <80.00%> (-0.25%)	⬇️
src/root.jl	79.16% <83.33%> (+1.96%)	⬆️
src/types.jl	92.40% <87.50%> (-1.11%)	⬇️
src/utils.jl	100.00% <100.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update a975f22...bd40294. Read the comment docs.

[Moelf]

@aminnj mentioned that when iterating over a (wide) tree, one should be lazy about unpacking baskets because there could be a continue branch in the event-loop that is executed 99.99% of the time (so you can ignore all other O(100) branches for almost all the time).

We don't have a tree iterator interface yet and this current design is hard to fit the bill, any suggestion would be welcomed.

[Moelf]

I combined the indexing and iteration interface, such that we support broadcast and map over a LazyBranch

[tamasgal]

Looks very good!